No Slide Title

Download Report

Transcript No Slide Title

Dominic F.G. Gallagher Thomas P. Felici

“propagating ideas”

Outline EME = “

E

igen

m

ode

E

xpansion”

•Introduction to the EME method - basic theory - stepped structures - smoothly varying - periodic •Why use EME?

Comparison to BPM and FDTD •Examples

“propagating ideas”

A

modes:

E

(

x

,

y

,

z

) 

e m

(

x

,

y

) 

e i

m z

B

The fields at AB of any solution of Maxwell’s Equations may be written as a superposition of the modes of cross section AB”.

“propagating ideas”

Basic Theory

propagation constant

E

(

H

(

x

,

x

,

y

,

y

,

z

)

z

)   

m

m

E H

m

(

x

,

m

(

x

,

y

)

y

)  (

c m

 .

e i

m

 (

c m

 .

e i

m z

z

c m

 .

e

i

m c m

 .

e

i

m z

)

z

) electric field magnetic field forward amplitude backward amplitude mode profiles •exact solution of Maxwell’s Equations •bi-directional

“propagating ideas”

•so far only z-invariant •what about a step change?

a m (+) a m (-) b b m m (+) (-)

Maxwell's Equations gives us continuity conditions for the fields, e.g. the tangential electric fields must be equal on each side of the interface

“propagating ideas”

•Applying continuity relationships, eg: LHS

k N

  1 

a

forward ( )

e k

i

k z

backward 

a k

(  )

e

i

k z

 .

E

(

a

k , t ) (

x

.

y

) 

k N

  1 

b k

(  )

e i

k z

forward 

b k

(  )

e

i

k z

 .

E

(b) k, t (

x

.

y

) backward RHS With orthogonality relationships and a little maths we get an expression of the form:

a

(  )

b

(  ) 

S J a

(

b

(  )  ) S J is the

scattering matrix

of the joint

“propagating ideas”

A

Straight Waveguide

B trivial - the

scattering matrix

is diagonal:

S AB

     

e i

 1

z

0 0  0

e i

 2

z

0  0 0

e i

 3

z

         

“propagating ideas”

Example: S-matrix decomposition of an MMI coupler

A B C D E

“propagating ideas”

A Evaluating S-matrix of device B C D E AB C DE ABC DE ABCDE

“propagating ideas”

Smoothly Varying Elements • Problem: modes are changing continuously along element • Thus each cross-section requires a large computational effort to locate the set of modes • This was the major hurdle that has in the past restricted application of EME • FIMMPROP (our implementation of EME) has tackled these problems, enabling EME to be used realistically for the first time even for 3D tapering structures.

“propagating ideas”

h n 0 Order (Staircase Approximation) h n Set of local modes computed at discrete positions along taper • Simple to implement • theoretically accurate as Nslice   • errors grow as Nslice   so practical limit on Nslice • can get spurious resonances between modes for long structures and small Nslice

“propagating ideas”

h n 1st Order (Linear Approximation) analytic integration h n Set of local modes computed at discrete positions along taper • More complex to implement • good accuracy for modest Nslice • errors  0 for modest Nslice • need only small number of modes

“propagating ideas”

Showing zero order versus first order result 1 0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0 0 100 200 300

taper length (um)

400 500

“propagating ideas”

Periodic Structures A B A B A B A B A B P1 P2 P1 P1 P1 P1 P2 P3 S • compute time  log(Nperiod) • i.e. almost as quick as a straight waveguide!

“propagating ideas”

transmission = (Sj) N periodic structure Sj Bends • bend is just periodic repeat of straight waveguide sections!

“propagating ideas”

Boundary Conditions •

PEC

&

PMC

(perfect electric/magnetic conductors) - useful for exploiting symmetries •

transparent

boundary conditions •

PML

’s - perfect matched layers (with PEC or PMC) • Transparent BC’s are naturally formed at input and output of EME computation • finding eigenmodes with true transparent boundary conditions leads to

leaky modes

- leaky modes cause problems with completeness of basis set.

• PML much better suited for finding modes for EME than leaky modes -

completeness

better achieved.

“propagating ideas”

PEC The Perfect Matched Layer (PML) d1 (real) d2=a+jb PEC waveguide core/cladding PML • Imaginary thickness of PML absorbs light

propagating

towards boundary • as much absorption as we wish with no reflection at cladding/PML interface!

• guided modes not absorbed at all - nice!

“propagating ideas”

Effect of PML on guided and radiating modes guided mode unbound mode PML core

“propagating ideas”

PML

PML’s with segmented waveguide

“propagating ideas”

Why Use EME?

EME Advantages 1. rigorous solution of Maxwell's Equations - rigorous as Nmode  infinity - large delta-n

“propagating ideas”

2. inherently bi-directional. - unconditionally stable since always express (outputs) = S.(inputs) - takes any number of reflections into account - not iterative - even highly resonant cavities - mirror coatings, multi-layer

“propagating ideas”

EME Advantages 3. The S-matrix technique provides the solution for all inputs!

- component-like framework where the S-matrix of one component may be re used in many different contexts.

Other methods: •Input 1  calculate  •Input 2  calculate  •Input 3  calculate  Result 1 Result 2 Result 3 EME/FIMMPROP: •Calculate S matrix •Input 1  Result 1 •Input 2  Result 2 •Input 3  Result 3

“propagating ideas”

EME Advantages 4. Hierarchical algorithm permits re-use, accelerating computation of sets of similar structures. When one part of a device is altered only the S-matrix of that part needs to be re-computed. • Initial evaluation time: ~ 2:54 m:s • change period - time: …… • change offset - time: ….

“propagating ideas”

Design Curve Generation

Traditional Tool: 5 mins 5 mins 5 mins EME/FIMMPROP: 5 mins

“propagating ideas”

3 mins 5 mins 5 mins 5 mins

EME Advantages 5. Wide-angle capability - wider angle - just add more modes - adapts to problem

“propagating ideas”

EME Advantages 6. The optical resolution and the structure resolution may be different.

- c.f. BPM (stability problems with non-uniform grid) • very thin layers - wide range of dimensions • no problem for EME/FIMMPROP algorithm does not need to discretise the structure

“propagating ideas”

Plasmonics Right: plasmon between silver plates EME is a rigorous Maxwell solution and can model many plasmonic devices (provided basis set is not too large).

“propagating ideas”

Why Use EME?

Disadvantages • Structures with very large cross-section are less suitable for the method since computational time typically scales in a cubic fashion with e.g. cross-section width.

• The algorithms are much more complex to write - for example it is very difficult to ensure that a mode has not been missed from the basis set.

• EME is not a "black box" technique - the operator must make some effort to understand the method to use it to his best advantage.

“propagating ideas”

Computation Time • Compare computation time with BPM and FDTD • Restrict our discussion to 2D - i.e. z and one lateral dimension • Both BPM and FDTD require a finite difference grid sampling the structure, this same grid used for optical field • EME does not need a structure grid (FMM Solver) • Equivalent of grid in EME is the number of modes • For straight wg, EME particularly efficient • Periodic section - logarithmic time • Arbitrary z-variations - all 3 methods have similar dependence with z complexity/resolution • In lateral dimension EME gets high spatial resolution almost for free. (c.f. BPM, non uniform grid problems…) • But lateral optical resolution - compute time  N 3 the

limiting factor

in EME

“propagating ideas”

Table1: Computation time – dependence on spatial resolution (2D simulation)

Straight waveguide – z dependence Near adiabatic structure – e.g. taper Periodic structure – z dependence Generic structure – z dependence Cross-section – structural resolution Cross-section – optical resolution BPM N 1 N 1 N 1 N 1 N 1 N 1 FDTD N 1 N 1 N 1 N 1 N 1 N 1 EME N 0 N 0  N 1 log(N) N 1  . N 1 ,  small N 3

“propagating ideas”

Memory Requirements • Very efficient as a function of z-resolution - N 0 scaling for straight or periodic

Table 2: Memory requirements – dependence on spatial resolution (2D simulation)

Straight waveguide – z dependence Periodic structure – z dependence Generic structure – z dependence Cross-section – structural resolution Cross-section – optical resolution BPM N 0  N 1 N 0  N 1 N 0  N 1 N 1 N 1 FDTD N N 1 N 1 N 1 N 1 1 EME N N 0  N 1 N 1 N 0 0  N N 1  N 2 1

“propagating ideas”

Applications • We present a variety of examples illustrating EME features

“propagating ideas”

SOI Waveguide Modes Ex Field

“propagating ideas”

Ey field

The MMI • MMI ideal for EME - inherently a modal phenomenon • 8 modes • Illustrate design scan - 100 simulations for price of two!

1 0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0 0

“propagating ideas”

0.5

1 1.5

input offset (um)

2 2.5

3

Tapered Fibre • 6 modes • “how long for 98% efficiency?” - ideal question for EME

“propagating ideas”

0.4

0.3

0.2

0.1

0 1 0.9

0.8

0.7

0.6

0.5

0 2000 4000 6000 taper length ( m m) 8000 10000 • 50 simulations in scan • 6.5s per simulation (in 3D!)

“propagating ideas”

Effective indices of first 5 modes 1.472

1.470

1.468

1.466

1.464

1.462

1.460

1.458

1.456

0 10 20 30 40 50 60 70 80 90 100 • Strong coupling occurs when the effective indices are close - telling us where the trouble spots of the device are.

• Powerful diagnostic - tells designer

where

to

Co-directional Coupler • remember logarithmic with N periods • very thin layer - no problem

“propagating ideas”

propagation at sub wavelength scales, including metal features

“propagating ideas”

Ring Resonator • Nmodes = 60 (for one ring) • Nmodes much higher here - wide angle propagation.

• BPM gives nonsense

“propagating ideas”

Photonic Crystal Design • Nmodes = 60 (for one ring) • Nmodes much higher here - wide angle propagation.

• BPM gives nonsense

“propagating ideas”

VCSEL Modelling top DBR mirror active layer lower DBR mirror • Resonance Condition

R top

.

R bot u

g u

R top R bot

“propagating ideas”

Silicon nanowires – high NA 2D 3D DBR Gratings EME  n FDTD Small NA BPM Fibre Taper Device length Comparison of computational domains of EME, BPM and Showing the domains of applicability of FDTD, BPM and EME to varying delta-n and device length.

“propagating ideas”

DBR Gratings FDTD EME BPM 3D 3D AWG cross-section size Showing the domains of applicability of FDTD, BPM and EME to varying numerical apperture and cross-section size.

“propagating ideas”

BPM – Capability Scores

Aspect

Speed Memory NA

Performance

FD-BPM scales linearly with area and can take fairly long steps in propagation direction -

Score/10

Usage scales linearly with c/s area Best with low NA simulations. Versions using Pade approximants can model a beam travelling at a large angle but still cannot deal well with light simultaneously travelling at a wide range of angles.

4 5 5 Delta-n Polarisation Lossy materials Reflections Non-linearity Dispersive Geometries Best with low delta-n simulations. Semi-vectorial versions work best. Still problems modelling mixed or rotating polarisation structures accurately Can model modest losses efficiently. Most versions cannot deal well with metals Some success in implementing reflecting/bi-directional BPM but generally avoided due to slow speed and stability problems FD-BPM can model non-linearity.

Being a frequency-domain algorithm this is easy The BPM grid allows diffuse structures to be modelled easily. Problems modelling non-rectangular structures accurately on the rectangular grid 7 3 9 10 7 ABCs PMLs available and work well 9

“propagating ideas”

Aspect

Speed Memory NA Delta-n Polarisation Lossy materials Reflections Non-linearity Dispersive Geometries ABCs FDTD – Capability Scores

Performance

Scales as V (device volume) but grid size is small so not as good as BPM or EME for long devices.

Scales as V (device volume) but grid size is small so not as good as BPM or EME for long devices.

Omni-directional algorithm is agnostic to direction of light – great when light is travelling in all directions Rigorous Maxwell solver, happy with high delta-n, but slows down somewhat with high index.

Rigorous Maxwell Solver is polarisation agnostic Can model even metals accurately with a fine enough grid and small modifications to the algorithm.

Yes – easy and stable even when there are many reflecting interfaces.

-

Score/10

10 9 10 10 Yes – non-linear algorithm relatively easy to do Have to approximate the dispersion spectrum with one or more Lorentizans but exact fit to the spectrum over a wide wavelength is difficult and the algorithm also slows down.

Fine rectangular grid can do arbitrary geometries easily, though there are problems approximating diagonal metal surfaces Yes – very effective and easy to use 9 7 8 9

“propagating ideas”

Aspect

Speed Memory NA Delta-n Polarisation Lossy materials Reflections Non-linearity Dispersive Geometries ABCs EME – Capability Scores

Performance

EME scales poorly with cross-section area – as A 3 (A is c/s area). However it can efficiently model very long structures especially if their cross-section changes only slowly or occasionally. Periodic structures scale as log(number of periods) – so can compute efficiently. S-matrix approach allows a set of similar simulations to be done very quickly – parts of previous calculation can be reused.

Memory increases at rate between A 2 and A 3 (A is c/s area), but very efficient for long or periodic devices.

Can model wide-angle beams by increasing the number of modes in the basis set at expense of speed and memory.

Rigorous Maxwell Solver can accurately model high delta-n Rigorous Maxwell Solver is polarisation agnostic Depends on mode solver used. -

Score/10

7 8 10 7 Yes – easy and stable even when there are many reflecting interfaces.

10 Difficult – have to iterate, and then only modest non-linearity levels will converge Being a frequency-domain algorithm this is easy Depends on the mode solver used. Can use different structure discretisations in different cross-sections, so solver can better adapt to the geometry. Depends on the mode solver used. E.g. a finite-difference solver can be readily constructed to implement PMLs. However, PML’s are more difficult to use with EME than with BPM or FDTD.

3 10 7 7

“propagating ideas”

Conclusions • Powerful compliment to BPM and FDTD • Exceedingly efficient/fast for wide range of examples • Rigorous Maxwell solver, bi-directional, wide angle • mode data provides important insight into workings of device.

“propagating ideas”